Investigation into relationships between grain size distribution characteristics and mechanical properties in large-scaled complex titanium alloy castings utilizing LASSO regression

The superimposition of multiple factors has presented a formidable challenge in establishing the salient features of grain size distribution for the large-scaled complex titanium alloy casting (LCTAC). In this study, an anatomical analysis within a LCTAC was conducted to acquire microstructure and mechanical properties distribution characteristics in typical zones. Through the progressive convergence, variable selection, and data dimensionality reduction techniques employed in the LASSO regression of microstructure characteristics and mechanical properties, the efficacious establishment of microstructural attributes has been accomplished. Consequently, a quantitative relationship between the microstructure and properties of the LCTAC has been formulated, predicated on the characteristics of grain size distribution. The findings unequivocally illustrate that the order of convergence for the microstructural parameters affecting the tensile strength of the LCTAC is as follows: expectation of grain size, mean grain size, full width at half maximum of grain size distribution, and the 99th percentile grain size. There exists a linear negative correlation between the 99th percentile grain size and tensile strength for the LCTAC, with a root mean square error in tensile strength prediction of less than 2%.